The AI Doppelgänger Dilemma: Cloned Voices in the Music Industry

The AI Doppelgänger Dilemma: Cloned Voices in the Music Industry Elizabeth Shields* ABSTRACT With the rise of generative artificial intelligence (AI), there has been an influx of “voice clones”—deep-learning algorithms that create synthetic speech to realistically mimic human voices. Celebrities and, in particular, music artists, have been subjected to the proliferation of AI voice clones on social media platforms like TikTok and streaming platforms such as Spotify. Despite music utilizing AI voice clones having amassed much popularity, this technology can be harmful and highly invasive to musicians whose livelihoods often depend on their distinct voices. While legal scholars have attempted to articulate various rights that could protect a person’s voice, individuals are largely left with minimal protection to prevent AI voice clones and have few options for redress. Some legal scholars suggest a variety of tort actions that could be applied in this context; however, torts such as the right of publicity, defamation, and false light ultimately fall short. This Note argues that a patchwork approach is necessary to regulate and combat the harms of AI voice clones, including action at the state and federal level, as well as self-regulation in the private sector by streaming platforms and musicians themselves. This approach, which includes input from all actors impacted by AI voice clones, should balance promoting creativity and continued development of AI while also protecting individuals’ interests in how their voice and likeness are used by others. * J.D. Candidate, Seattle University School of Law, 2025. 761 762 Seattle University Law Review [Vol. 48:761 CONTENTS INTRODUCTION ......................................................................................763 I. CURRENT SOLUTIONS IN TORT LAW ..................................................765 A. Right of Publicity—Introduction and Relevant Case Law .............766 B. Benefits and Drawbacks in Applying the Right of Publicity to AI Voice Clones ...............................................................768 C. Other Torts: Defamation and False Light .....................................769 D. Strength Assessment of Tort Claims ..............................................770 II. PROPOSED SOLUTIONS—PRIVATE SPHERE .......................................771 A. Royalties Approach ........................................................................771 B. Terms and Conditions, Platform Rules, and User Guidelines ..................................................................................772 C. Licensing Agreements ....................................................................775 D. Strength Assessment of Private Sphere Solutions ..........................776 III. PROPOSED SOLUTIONS—PUBLIC SPHERE .......................................776 A. No Fakes Act Overview..................................................................777 B. No Fakes Act Efficacy ....................................................................778 1. Benefits......................................................................................778 2. Limitation: Lack of Platform Accountability ............................779 3. Limitation: Enforcement Challenges.........................................779 4. Limitation: Consent and First Amendment Concerns ...............780 5. Limitation: Impact on Existing Laws ........................................781 C. Other Approaches: States and the European Union......................781 CONCLUSION..........................................................................................784 2025] The AI Doppelgänger Dilemma 763 INTRODUCTION In April 2023, an anonymous TikToker named Ghostwriter977 posted a song called “Heart on My Sleeve” featuring music artists Drake and The Weeknd (referred to collectively as “Drake”).1 The song went viral and amassed millions of streams on YouTube, Spotify, and Apple Music.2 However, Drake did not actually contribute to the song; instead, Ghostwriter977 used an AI model to create a clone of Drake’s voice that was indistinguishable from the real Drake.3 AI voice clones, also referred to as “verbal deepfakes,” are produced through deep-learning algorithms that create synthetic speech to realistically mimic human voices.4 Products like Amazon Echo or Google Home have long offered basic text-to-speech capabilities,5 but in recent years, numerous AI companies have significantly advanced the quality of computer voice generators.6 Unlike previous methods that relied on assembling fragments of existing recordings,7 voice-cloning products employ predictive models built with generative AI to reproduce English words in a more natural and human-like manner, distinguishing them from other computer voices, such as Apple’s Siri.8 Despite the viral popularity of “Heart on My Sleeve,” the proliferation of AI voice clones presents a myriad of concerns for artists, streaming platforms, and the general public. Music artists’ market value depends on the integrity of their voices and the quality of their songs, with vocals considered a “sacred ingredient of [musicians’] art and profession.”9 Fans expect consistent branding,10 which is put at risk when individuals like Ghostwriter977 use doppelganger voice clones, rendering it difficult for the public to discern whether a song is created by the real Drake or AIDrake. 1. Mia Sato, Drake’s AI Clone is Here—and Drake Might Not Be Able to Stop Him, VERGE (May 1, 2023), https://www.theverge.com/2023/5/1/23703087/ai-drake-the-weeknd-music-copyright-legalbattle-right-of-publicity [https://perma.cc/4E3S-MZZX]. 2. Id. 3. Id. 4. See Bernard Marr, How AI Creates Synthetic Speech, FORBES (Nov. 2, 2021), https://www.forbes.com/sites/bernardmarr/2021/11/02/how-ai-creates-synthetic-speech/. 5. Bernard Marr, The Difference Between Generative AI and Traditional AI: An Easy Explanation for Anyone, FORBES (Aug. 23, 2023), https://www.forbes.com/sites/bernardmarr/2023/07/24/thedifference-between-generative-ai-and-traditional-ai-an-easy-explanation-for-anyone/?sh=23a730b1508a. 6. Id. 7. What Is Text to Speech & How Does It Work?, WELLSAID (Nov. 3, 2021), https://wellsaidlabs.com/blog/what-is-text-to-speech/ [https://perma.cc/JX62-7YUD]. 8. Id. 9. Martina, AI vs. Music Industry: The Rise of AI Voice Cloning, IMUSICIAN: BLOG (Oct. 12, 2023), https://imusician.pro/en/resources/blog/the-rise-of-ai-voice-cloning [https://perma.cc/BM9TUJUU]. 10. See id. 764 Seattle University Law Review [Vol. 48:761 Additionally, the spread of AI voice clones can be framed as a tragedy of the commons problem.11 Within days of the removal of “Heart on My Sleeve,” another TikTok account, Ghostwrider977 (distinguished from the original “Ghostwriter977” account), released an original song using AI voice clones of music artists Rihanna and Bad Bunny.12 Posts containing voice clones of famous musicians have since flooded the platform, sparking Unive (...truncated)